Exploiting Chinese character models to improve speech recognition performance

The Chinese language is based on characters which are syllabic in nature. Since languages have syllabotactic rules which govern the construction of syllables and their allowed sequences, Chinese character sequence models can be used as a first level approximation of allowed syllable sequences. N-gram character sequence models were trained on 4.3 billion characters. Characters are used as a first level recognition unit with multiple pronunciations per character. For comparison the CU-HTK Mandarin word based system was used to recognize words which were then converted to character sequences. The character only system error rates for one best recognition were slightly worse than word based character recognition. However combining the two systems using log-linear combination gives better results than either system separately. An equally weighted combination gave consistent CER gains of 0.1-0.2% absolute over the word based standard system. Copyright © 2009 ISCA.

The development of turbine exit flow in a swan-necked inter-stage diffuser

This paper describes both the migration and dissipation of flow phenomena downstream of a transonic high-pressure turbine stage. The geometry of the HP stage exit duct considered is a swan-necked diffuser similar to those likely to be used in future engine designs. The paper contains results both from an experimental programme in a turbine test facility and from numerical predictions. Experimental data was acquired using three fast-response aerodynamic probes capable of measuring Mach number, whirl angle, pitch angle, total pressure and static pressure. The probes were used to make time-resolved area traverses at two axial locations downstream of the rotor trailing edge. A 3D time-unsteady viscous Navier-Stokes solver was used for the numerical predictions. The unsteady exit flow from a turbine stage is formed from rotordependent phenomena (such as the rotor wake, the rotor trailing edge recompression shock, the tip-leakage flow and the hub secondary flow) and vane-rotor interaction dependant phenomena. This paper describes the time-resolved behaviour and three-dimensional migration paths of both of these phenomena as they convect downstream. It is shown that the inlet flow to a downstream vane is dominated by two corotating vortices, the first caused by the rotor tip-leakage flow and the second by the rotor hub secondary flow. At the inlet plane of the downstream vane the wake is extremely weak and the radial pressure gradient is shown to have caused the majority of the high loss wake fluid to be located between the mid-height of the passage and the casing wall. The structure of the flow indicates that between a high pressure stage and a downstream vane simple two-dimensional blade row interaction does not occur. The results presented in this paper indicate that the presence of an upstream stage is likely to significantly alter the structure of the secondary flow within a downstream vane. The paper also shows that vane-rotor interaction within the upstream stage causes a 10° circumferential variation in the inlet flow angle of the 2nd stage vane.

Syllable language models for Mandarin speech recognition: exploiting character language models.

Mandarin Chinese is based on characters which are syllabic in nature and morphological in meaning. All spoken languages have syllabiotactic rules which govern the construction of syllables and their allowed sequences. These constraints are not as restrictive as those learned from word sequences, but they can provide additional useful linguistic information. Hence, it is possible to improve speech recognition performance by appropriately combining these two types of constraints. For the Chinese language considered in this paper, character level language models (LMs) can be used as a first level approximation to allowed syllable sequences. To test this idea, word and character level n-gram LMs were trained on 2.8 billion words (equivalent to 4.3 billion characters) of texts from a wide collection of text sources. Both hypothesis and model based combination techniques were investigated to combine word and character level LMs. Significant character error rate reductions up to 7.3% relative were obtained on a state-of-the-art Mandarin Chinese broadcast audio recognition task using an adapted history dependent multi-level LM that performs a log-linearly combination of character and word level LMs. This supports the hypothesis that character or syllable sequence models are useful for improving Mandarin speech recognition performance.